The Cambrian Explosion

"Trace-fossil information strongly supports the Cambrian explosion, but allows for a short time of phylogenetic fuse during the terminal Ediacaran–Fortunian." — M. Gabriela Mángano and Luis Buatois, 2014, from the one line of evidence that needs no exceptional preservation.¹

The "clumpiness" of animal form "is a consequence of the extinction of phylogenetic intermediates, indicating that the original distribution of morphologies was more homogeneous." — Deline and colleagues, 2018, in the strongest published result against this article's argument.²

"Pikaia is not an annelid worm.  It is a chordate, a member of our own phylum." — Stephen Jay Gould, 1989.  Those are the pages a creationist paper cites for the claim that Pikaia sits between two phyla.³

Summary

In rock assigned to the first part of the Cambrian period, animals with hard parts, legs, eyes, guts and shells appear in large numbers.  In the rock below, they do not.  That contrast is the Cambrian explosion, and it has been argued about since Darwin devoted a section of the Origin to it.

This article asks three questions.  Is the pattern real, or is it an artefact of when animals became easy to fossilise?  How big is it, measured rather than asserted?  And which of the claims made about it, by creationists, by intelligent-design writers and by their critics, survive a check against the sources?

The short answers.  The pattern is real.  Its magnitude is not measurable as precisely as anyone on either side pretends.  And most of the specific numbers this site's own tradition has quoted about it are wrong.

That is a mixed verdict, and more of it goes against this site's own tradition than for it.  The useful part is knowing which pieces to keep.

What this article concedes before it argues anything

A skeptical reader should get the bad news first, because otherwise they will find it themselves and stop reading.

The rocks below the Cambrian are not empty.  They hold a large, worldwide, well-sampled assemblage of macroscopic organisms, the Ediacaran biota, and at least one of them, Kimberella, is an animal with bilateral symmetry found alongside the scratch marks it made while feeding.⁴  Anyone who says animals appear from nothing is describing a record that does not exist.

The strongest published measurement of animal form says the gaps were made afterwards.  Deline and colleagues built a morphospace across the whole animal kingdom, dropped fossils and reconstructed ancestors into it, and concluded that the clumpiness of animal form is a consequence of the extinction of intermediates, and that arthropods, chordates, annelids, echinoderms and molluscs have kept expanding the range of form they occupy ever since the Cambrian.²  That is a direct measurement, it is recent, and it says the two things this article's argument least wants to hear.

A mass extinction sits inside the Cambrian, at almost exactly the point where this article's own headline result stops.  Zhuravlev and Wood place the Sinsk Event, a global episode of shallow-marine anoxia, early in Cambrian Stage 4.⁵  This article's own computation finds that phylum-rank first appearances stop at the top of Stage 3, immediately below it.  An extinction can end a wave of first appearances by destroying the fauna producing them.  That confound is not settled here, and it is printed in Part 3 rather than buried.

The best young-earth paper on transitional fossils contains a citation error, and this article prints the correction.  Kurt Wise's 1995 concession is the most-quoted creationist statement in this series.  Its only phylum-level example is Pikaia, described as sitting "between echinoderms and chordates" and sourced to three pages of Gould.⁶  Those pages say Pikaia is a chordate.  The claim is not on the pages cited, and Pikaia has moved twice since, both times further inside the chordate stem lineage.³⁷  Part 8 says so plainly.

And the taxonomic ranks this argument is usually phrased in are partly a matter of assignment practice, not of rock.  A "phylum" is a category a taxonomist assigns.  When a strange Cambrian animal is reclassified from "its own phylum" to "a stem group between two living phyla," the count of Cambrian phyla falls without a single fossil changing.  Zhang and Shu name Anomalocaris and Opabinia as exactly that case.⁸  Any claim of the form "N phyla appear at once" is therefore partly a claim about how paleontologists file things.  The series flagship works this objection through at length, and this article does not pretend to have escaped it.⁹

None of this depends on radiometric dating

Radiometric dating is a separate subject with separate articles on this site.  Nothing here rests on it, and no argument here rests on a duration in years.

Every "million years ago" figure below is a label for a slice of rock under discussion, the interval whose contents are being described, and never this article's evidence.  The argument runs on relative order.  Within one section that is superposition: younger rock sits on older rock.  Between sections and between continents it is the matching of rock by its fossils into one global sequence of stages, which is how a worldwide database can file a Siberian fossil and a Chinese fossil in the same stage.  Young-earth geology accepts both.  Kurt Wise and Donna Richardson, testing the stages against the same database this article uses, concluded that "the global biostratigraphic column seems to be real."¹⁰

Stated without any dates at all, the three load-bearing observations are these.  Burrows and trails become far more varied and far deeper across the Ediacaran–Cambrian boundary.  Fossil-bearing rock immediately below that boundary contains preserved algae and no animals of Cambrian type.  And in the animal record above it, first appearances at high taxonomic rank are concentrated in the lower part of the Cambrian and then stop, while first appearances at low rank keep running.  Every one of those is a statement about what lies above what.

Where duration is discussed, in Part 4, the numbers are quoted as the mainstream figures being argued over.  They are used to show that one side's stated duration misreports the mainstream's own timescale, which is a comparison between published claims, not a dating argument of this site's own.

What was computed for this article

Quoting somebody's summary of a number is weaker than computing the number.  Two analyses were run against the Paleobiology Database, which is open, free, and maintained by working paleontologists.¹¹

• 48,500 occurrence records of animals from Ediacaran and Cambrian rock, binned by stage under a strict age-containment rule.¹¹
• Every animal phylum, class, order and family in the database with at least one occurrence, binned by the period of its first appearance, across all of geological time.¹¹

The scripts are published beside this article and anyone can re-run them.¹²

Two warnings travel with every number produced this way.  The database records published occurrences, so it measures the paleontological literature and not the rocks directly.  And sampling across the Cambrian is wildly uneven: Cambrian Stage 3 has twenty-seven times the well-dated collections of the Fortunian.¹¹  Part 3 prints the sampling table beside the result table, because printing one without the other would be cherry-picking.

Part 1: The trace fossils, which are the best evidence in this dispute

Verdict: the strongest single line of evidence that the explosion is a real event, and it is barely used on either side of the popular argument.

Every other section of this article has to survive the same objection.  It runs like this: the Cambrian is not when animals appeared, it is when animals became easy to fossilise.  Shells, spines and hardened cuticle preserve.  Naked worms mostly do not.  If a dozen lineages hardened up over a short interval, a dozen lineages would appear over a short interval, whether or not anything happened to them evolutionarily.

A burrow answers that objection by not being a body.

Why a burrow beats a shell

A trace fossil is a record of behaviour preserved in sediment: a burrow, a trail, a scratch mark, a trackway, a resting impression.  Making one requires no skeleton.  Preserving one requires no anoxic bottom water, no Lagerstätte, and no unusual ocean chemistry.  It requires ordinary sediment and an animal that disturbed it.

Mángano and Buatois state the advantage directly: the record of animal–sediment interactions across the Ediacaran–Cambrian boundary "is far more continuous than the body-fossil record, particularly with respect to that of soft-bodied organisms."¹

There is a second advantage, and it is the one that matters for a body-plan argument.  Ichnologists count two different things.  Ichnodiversity is the number of distinct named trace types, the richness measure.  Ichnodisparity is the number of distinct architectural designs: how many fundamentally different construction plans the traces represent.  Ichnodisparity is the behavioural analogue of a body plan.  A creature that can dig a deep vertical shaft with a lined wall is built differently from one that grazes a mat surface, and the difference shows in the sediment whether or not either animal ever fossilises.

The numbers, from a global compilation

Mángano and Buatois assembled a worldwide database of Ediacaran and Cambrian trace fossils and tabulated it by time slice.  Their Table 1 is the most useful single table in this literature.¹

Read the Fortunian column.  Named trace types go from 7 to 42, and 33 of the 42 are new.  Architectural designs, the disparity measure, go from 5 to 22.  Maximum burrowing depth goes from one centimetre to eight, and to a full metre by Stage 2.  The bioturbation index, which measures how thoroughly the seafloor has been churned, goes from 0 to 3 and then to 6, its maximum.

That is the Cambrian explosion measured in behaviour rather than in bodies, and it gives the same answer.  No shell is involved anywhere in that table.  The two ichnologists who built it draw the conclusion themselves: trace-fossil information "strongly supports the Cambrian explosion."¹

The base of the Cambrian is itself a burrow

Almost no popular account mentions this, and a reader needs it before anything else.

The Cambrian period does not begin at a shell.  Its base is fixed at a Global Boundary Stratotype Section and Point at Fortune Head, Newfoundland, defined by the first appearance of a trace fossil: the branching burrow Treptichnus pedum.  Erwin and colleagues state what that burrow implies: "the first appearance of bilaterian animals with the ability to make complex burrows both horizontally and vertically."⁴

Two consequences follow, and they point in opposite directions.  Both belong in an honest account.

Against this article's framing: the boundary was drawn at evidence of bilaterian animal activity.  So "bilaterians appear at the base of the Cambrian" is partly true by construction.  Move the golden spike and the sentence changes.

Against the preservation explanation: the same fact means the base of the Cambrian is not a preservation threshold.  Nothing about Treptichnus requires a hard part or an unusual burial.  Whatever made it was already a burrowing bilaterian, and nothing burrowing in that way is known much below it.

What is missing from the Ediacaran, itemised

The negative list in Mángano and Buatois is more useful than the positive one.¹

• Arthropod traces are "strikingly absent." No trackways, no Rusophycus resting impressions, no Diplichnites.
• No vertical burrows in higher-energy sediments, the structures a suspension feeder makes.
• No sophisticated feeding programmes. Nothing shows systematic U-turns, avoidance of one's own trail, or tracking alongside a previous pass. Structures once read as guided meandering grazing trails have been re-examined and are now taken to be body fossils, because the meanders are not there.
• The bioturbation index is zero through most of the Ediacaran and occasionally one near its top. The seafloor was a microbial mat, not a churned bed.
• Simple horizontal grazing trails on mats are 59% of all recorded Ediacaran cases, and 52% of Ediacaran trace occurrences are monospecific: one trace type on the bedding plane and nothing else. Their comparison is that this stands in sharp contrast to Phanerozoic trace assemblages.

What the Ediacaran traces do establish, which is a concession

The same authors are equally clear on the other side, and this article prints it because they made the point against their own convenience.

Apart from two problematic forms, every trace type present in the Ediacaran continues into the Phanerozoic, where bilaterian animals are known to make them.  Mángano and Buatois argue it is ad hoc to assume that identical structures in Ediacaran rock were made by something other than bilaterians.¹  So the makers of the Ediacaran traces were probably bilaterian animals, and the change across the boundary is a change in what those animals could do, not the sudden arrival of the first ones.

They also decline to count evidence in their own favour.  They attribute the fan-shaped paired scratch marks to Kimberella on ichnological grounds, the animal sitting at the apex of the fan, while noting that its phylogenetic position remains disputed.  And they leave Epibaion, the resting and locomotion traces associated with Dickinsonia and Yorgia, out of their table altogether, because those animals' affinities are uncertain and passive transport has been proposed for the marks.¹  A source that excludes its own best examples is a source worth citing.

The result that constrains everything else: traces run ahead of bodies

Daley and colleagues assembled the arthropod case, and its structure matters more than any single number.¹³

• Trilobites, arthropods with mineralised exoskeletons, first appear as body fossils at about 521 Ma.
• Burgess Shale-type deposits preserving soft-bodied arthropod relatives, including Anomalocaris and its kin, begin at about 518 Ma.
• Trace fossils attributed to arthropods appear at about 537 Ma: Cruziana and Rusophycus, the plough marks and resting impressions.

So the traces run about two stages ahead of the earliest arthropod body fossils: arthropod traces in the Fortunian, the lowest Cambrian stage, and trilobite bodies at the base of Stage 3.  On the mainstream timescale that gap is roughly sixteen million years.  And the exceptionally preserved soft-bodied forms turn up after the armoured ones, not before.  Daley and colleagues call the interval between them a ghost lineage that needs constraining, and the traces are what constrains it.

Both sides should be uncomfortable with that, and this article says so.

For the mainstream account, the soft-bodied stem lineage that ought to document the assembly of the arthropod body plan is preserved later than the finished product.  The evidence arrives in the wrong order relative to the story it is asked to tell.

For this article's side, the trace record shows arthropod-grade animals walking and resting on sediment two stages below the first preserved arthropod body.  That is a documented, non-hypothetical case of a group existing well before its body-fossil first appearance, which is precisely the mechanism the preservation objection appeals to.  Anyone who argues that first appearance equals origination has just been shown a counterexample from the very group the argument is usually built on.

The limits of the trace evidence, printed

Four, and none of them is fatal.

Trace names are not animal names.  One animal makes several kinds of trace depending on what it is doing; several unrelated animals make traces that look identical.  Ichnotaxonomy is a classification of structures, not of organisms, so a jump in ichnodiversity is not a headcount of new species.  Ichnodisparity is the better measure for exactly this reason, and it is the one that goes 5 to 22.

The trace compilation is a literature compilation too.  Mángano and Buatois assembled published occurrences, so their table inherits whatever bias the publication record carries.  What blunts this is that the Ediacaran trace record has been intensively searched, precisely because finding a complex Precambrian burrow would be a career-making result.

Some of the Ediacaran shortfall is a facies effect.  Matground-dominated seafloors preserve and permit different behaviours than churned ones, so part of the contrast is environmental rather than evolutionary.  That is real, but it is also circular as a full explanation, because the matgrounds persisted because nothing was churning them, and the churning is the thing to be explained.

The depths and bioturbation grades in the table are maximums, not typical values.  A maximum is set by the deepest burrow or the most churned bed anywhere in the compilation.  Typical early Cambrian burrowing was shallow and patchy.  In the trilobite-bearing lower Cambrian carbonates of the western United States, burrows averaged 2 to 3 centimetres deep,¹⁴ and lower Cambrian strata are most often at the two lowest grades of disturbed bedding.¹⁵  The argument of this Part rests on the new designs, which one well-made burrow establishes.  It does not rest on a thoroughly churned seafloor, which the lower Cambrian did not have.

Part 2: What lies under the Cambrian

Verdict: not an empty world, not an ancestral fauna, and both overstatements should be abandoned.

The single strongest response to "animals appear without ancestors" is that the rock immediately below the Cambrian is richly fossiliferous.  The single strongest response to "the Ediacaran supplies the ancestors" is that almost none of it can be placed in the animal kingdom, and that most of it dies out.

The three assemblages

Erwin and colleagues divide the Ediacaran macrofossil record into three successive assemblages, and the division is standard.⁴

Marshall adds the front end: the Ediacaran biota appears shortly after the Gaskiers glaciation, and near the top of the Ediacaran the first lightly skeletonised forms turn up: the cone-shaped Cloudina and the more complex Namacalathus.¹⁶

And Erwin and colleagues give the exit line as flatly as it can be given: "Ediacara macrofossils are not known from the Phanerozoic and evidently went extinct by the Cambrian."⁴

That sentence cuts both ways.  It means the Ediacaran biota is not a smooth ancestral fauna grading into the Cambrian one; most of it disappears.  It also means the Cambrian is not the first appearance of large complex organisms on earth.

How much of the Ediacaran is animal? Two clades out of nine

This is the number to lead with, and it comes from the mainstream side.

Erwin and colleagues recognise nine likely clades of Ediacaran organisms.  Of those, only two can confidently be assigned to the crown Metazoa: the kimberellomorphs, and less securely the dickinsoniomorphs, with putative sponges set aside separately.⁴

That is a genuinely mixed result and both popular readings of it are wrong.  A creationist who says the Ediacarans are not animals at all is contradicted by at least one of them.  A textbook that says the Ediacaran supplies the ancestors is overstating a record in which seven of nine clades cannot be placed in the animal kingdom with confidence.

Kimberella: a body plus a behaviour

Kimberella is centimetre-sized and bilaterally symmetrical, with a crenulated margin read as a frill around a muscular foot, and a proboscis.  Erwin and colleagues call these fossils bilaterians and possible molluscs, and note that they are commonly found together with radiating scratch traces, the ichnogenus Radulichnus, that appear to record feeding on microbial mats.⁴

The combination is what makes the case.  A body fossil on its own can be argued about indefinitely; every unusual flattened impression attracts a dozen readings.  A body fossil found at the apex of a fan of paired scratch marks, of the kind a rasping mouthpart leaves on a mat, is a body and a behaviour.  The behaviour requires a muscular anterior feeding structure and directed forward movement.  That is bilaterian-grade organisation whatever one decides to call the animal.

Verdict: Kimberella is a pre-Cambrian animal with bilaterian-grade organisation, and "there are no animals before the Cambrian" is dead on this fossil alone.

The honest limits: the mollusc assignment is an interpretation, and "possible molluscs" is Erwin's own hedge.  And Kimberella is one genus, not a fauna.  It does not supply an ancestor for anything in particular.

Dickinsonia: the chemical result, with its unresolved objection

Dickinsonia is a flat, ribbed, oval impression, and for decades it was read as a worm, a jellyfish, a lichen, a giant single-celled protist, or a failed experiment belonging to no living kingdom.  In 2018 Bobrovskiy and colleagues settled a large part of the question by chemistry rather than by anatomy.

The method is worth explaining, because it is what makes the result hard to dismiss.  Living organisms build membranes out of sterols.  Sterols degrade, over geological time, into steranes and aromatic steroids that survive in rock.  Their carbon-number distribution differs by kingdom: animals are dominated by C27 cholesteroids, fungi by C28 ergosteroids, green algae and plants by C29 stigmasteroids.  Bobrovskiy and colleagues extracted these compounds from the organic films that are the fossils, at the Lyamtsa and Zimnie Gory localities on the White Sea coast of Russia.

The numbers.  Sediment immediately above and below the fossils runs 10.6–11.9% cholesteroids, 13.4–16.8% ergosteroids and 71.3–76.0% stigmasteroids, an algal background.  The largest Dickinsonia specimen runs 93% cholesteroids, 1.8% ergosteroids and 5.2% stigmasteroids.  Correcting for the surrounding algal signal, the authors compute that the living animal's sterols were at least 99.7% cholesteroids.¹⁷

Why that kills the lichen reading specifically.  Lichen-forming fungi make ergosteroids as their major sterols even when they host algal partners.  Dickinsonia has at most a fifth of a percent.  The other main alternative, a giant rhizarian protist, produces mixtures spanning wide ranges of all three and may make C30 sterols besides.  Neither profile matches.

The internal control that makes the contamination objection hard.  Cholesteroid abundance rises from 84.8% in the smaller specimens to 93.0% in the larger ones, while the other two fractions stay put.  A contaminant should not scale with the size of the fossil it is contaminating.

And the published doubt, which is not a creationist invention.  Summons and Erwin's commentary in the same issue of Science raises a stereochemistry problem: steroids in sedimentary rock are normally mixtures of stereoisomers produced by burial and heating, yet the Dickinsonia cholesteroids are dominated by a single form.  This article could not obtain that commentary, which is paywalled, so this paragraph rests on secondary reports and is flagged as unverified.¹⁸  Note that Bobrovskiy and colleagues address the same isomer ratio and read it as the signature of anaerobic microbial decay of a carcass, which would be evidence for a decaying animal body rather than against one.

Verdict: probably an animal, on evidence a skeptic and a creationist should both find hard to dismiss, with one analytical objection this article has not been able to check.  What Dickinsonia is not is an ancestor of anything Cambrian.  Erwin and colleagues suggest a stem placozoan or stem eumetazoan, which is below the bilaterians, and the biomarker paper does not resolve the position either.

Ikaria: a millimetre-scale bilaterian tied to its own trail

Evans and colleagues described Ikaria wariootia from the Ediacara Member of the Rawnsley Quartzite in South Australia: 108 specimens on a single bed surface plus 19 more from loose blocks, preserved as impressions.  It is tiny, from 1.9 to 6.7 mm along its long axis.¹⁹

The argument they build is indirect and should be stated honestly.  A small leveed horizontal burrow called Helminthoidichnites is abundant in the same beds and occurs more than 100 m of section below the first appearance of Kimberella.  Its geometry requires an animal that displaced sediment while moving under thin sand bodies, which requires a front end distinct from a back end and a body cavity to move against.  Evans and colleagues show that Ikaria's size-frequency distribution cannot be distinguished from that of the burrows, and that no other body fossil in the deposit has the right combination of size and front-to-back differentiation.  They conclude it is the trace-maker.

Verdict: a good inference, not a proof.  Matching size distributions is circumstantial, and the paper's own phrasing is that Ikaria is the interpreted progenitor of the traces.  What is not in doubt is the burrows themselves, and the burrows are the harder evidence.  A small bilaterian animal was moving through sediment well below the Cambrian boundary, whatever it looked like.

The thing that is missing from the Ediacaran, and it is the mainstream's own point

This is the most useful paragraph in this article for its argument, and every word of it comes from mainstream sources.

Daley and colleagues argue that a deep Precambrian root for the arthropods is ruled out by the Ediacaran preservation windows themselves.  Their reasoning is that Burgess Shale-type deposits exist in the latest Ediacaran, the Miaohe biota at about 550 Ma among them, and are abundantly fossiliferous.  What they contain is algae.  Their words: such deposits "are abundantly fossiliferous with algae but completely lack animals."¹³  Ediacaran phosphate deposits such as Doushantuo at about 560 Ma, which preserve cellular and even sub-cellular detail, likewise contain no arthropods.  These are exactly the preservation modes that capture soft-bodied arthropod relatives in the Cambrian.  They are not empty.  They simply have no animals of that kind in them.

Erwin and colleagues make the matching point about skeletons: Ediacaran phosphate deposits are common and contain no small shelly fossils, which suggests to them that bilaterian groups acquired skeletons during the Cambrian rather than before it.⁴

That removes the easiest evolutionary answer.  The absence of Cambrian-type animals below the Cambrian is not simply the absence of the right rocks or the right preservation mode.  The right rocks exist, the right preservation mode operates in them, and what comes out is algae.

But state the limit too, because it is a real one.  Daley's argument is about arthropods and about macroscopic soft-bodied preservation.  Erwin and colleagues and Zhang and Shu both argue that the earliest bilaterians were very small, smaller than these windows capture, and too small to leave a preservable trace.  Zhang and Shu say the absence of convincing metazoan fossils earlier than about 560 Ma strongly implies a cryptic early evolution of animals.⁸

That is a hypothesis, not an observation, and this article labels it as one.  It has not been falsified and it may well be true; Ikaria at two millimetres is roughly the size the hypothesis calls for.  But the same standard has to be applied in both directions.  An argument from the absence of Precambrian ancestors is also an argument from what has not been found.

Part 3: What appears and when, computed from 48,500 records

Verdict: first appearances are strongly front-loaded by taxonomic rank, the effect is large, and it is a statement about first appearances and nothing else.

Almost every published version of this argument quotes somebody's summary.  This one measures it.

The method, and a correction that changes the answer

Every animal occurrence in the Paleobiology Database from Ediacaran and Cambrian rock was downloaded, 48,500 records, together with every animal phylum, class, order and family in the database and its recorded first appearance.¹¹¹¹

One methodological point has to come first, because getting it wrong produces a result that flatters this article's argument and is false.

Each occurrence in the database carries an age bracket, not a date.  A record given as 538.8–520 Ma spans four Cambrian stages.  The database's own binning field assigns such a record to whichever single stage covers the largest slice of the bracket, which for that example is the Fortunian, the oldest one it touches.  Run naively, this produces the headline "the first arthropods are Fortunian, from the Campito Formation of California."  Those nine trilobite records are in fact from two stages higher.  Taking the database's binning at face value systematically backdates poorly constrained records.

So every table below uses a stricter rule: an occurrence counts toward a stage only when both ends of its age bracket fall inside that stage.  Records that straddle a boundary are reported separately rather than pushed into the oldest stage they touch.  That rule is conservative in the direction that hurts this article: it moves first appearances later, never earlier.

The result: phylum-rank first appearances stop

Under the strict rule, the cumulative count of phylum-rank groups with a first appearance at or below each stage runs:

Not one phylum-rank first appearance falls in Cambrian Stage 4 or in any later Cambrian stage.  Twenty-five accumulate from the Ediacaran to the top of Stage 3, and then the count stops dead for the remaining seven stages of the Cambrian.  The one group that cannot be checked is Tardigrada, whose single Cambrian occurrence is dated only to "Middle Cambrian" and cannot be pinned to a stage.

Two clarifications before the number is quoted anywhere.  This is a count of the database's own phylum-rank groups, not of textbook phyla, so it includes entries such as Problematica and the disputed Ediacaran groups; the caveats at the end of this part say why that list should never be quoted raw.  And "the count stops" means it stops inside the Cambrian.  Ten phyla first appear in the record long after the Cambrian ends, and that is dealt with below rather than hidden.

The order of arrival is worth having, because it is not random.  Annelids, cnidarians, sponges and the Ediacaran problematic groups are already present in Ediacaran rock.  Molluscs, chaetognaths and scalidophorans arrive in the Fortunian.  Arthropods, brachiopods, chordates, onychophorans and priapulids arrive in Stage 2.  Bryozoans, ctenophores, echinoderms, entoprocts, hemichordates, kinorhynchs and flatworms arrive in Stage 3.  Then nothing.

The rank ladder, across all of geological time

This is the disparity-before-diversity claim turned into a measurement.  Every animal phylum, class, order and family in the database that has at least one occurrence, binned by the period in which it first appears.¹¹

The ladder is monotone and steep.  Half of all fossilised animal phyla and half of all fossilised animal classes make their first appearance inside one period.  Only 13% of orders and 6.6% of families do.

And there is a built-in control against the obvious objection.  If high-rank front-loading were purely an artefact of taxonomists giving grand names to the oldest thing they find, the Ordovician should behave like a weaker Cambrian at every rank.  It does not.  The Ordovician produces more first-appearing orders than the Cambrian, 145 against 139, and more families, 755 against 609.  What it does not produce is phyla and classes: one and seventeen.  The Ordovician out-competes the Cambrian at low rank and is crushed by it at high rank.  That is a shape a pure naming artefact does not predict.

The same ladder holds inside the Cambrian, across the line at the top of Stage 3:

Phyla go 15 to 1 across that line, although the second interval holds more than twice as many stages.  Families go 359 to 250, a modest drop.  The higher the rank, the more sharply it is concentrated in the first three stages.

One bookkeeping point, because the two tables in this part look as though they disagree.  This table is built from the database's own recorded first-appearance ages;  the cumulative table above is built from occurrences under the strict-containment rule.  The Cambrian's sixteenth phylum-rank group is Tardigrada, whose single Cambrian occurrence is dated only to "Middle Cambrian" and therefore cannot be pinned to a stage at all.  It is the 1 in the second column here, and it is the group the strict rule drops.  No phylum-rank first appearance can be placed in Stage 4 or later by the strict rule, and the one record that might belong there cannot be placed anywhere.

Genus counts make the contrast sharper still.  By the top of Stage 3, 100% of the Cambrian's phylum-rank first appearances and 76% of its class-rank first appearances have happened, against 47% of its genus first appearances.  Genera keep arriving at between 87 and 261 per stage right through to the Ordovician boundary.  High ranks finish early; low ranks keep running.

Now the sampling table, which must be printed beside the result

Only strictly stage-contained records appear here; 28,052 of the 48,500 occurrences straddle a stage boundary and are excluded.¹¹

Cambrian Stage 3 has twenty-seven times the well-dated collections of the Fortunian, and on the database's chart sixty times the occurrences per million years.  That is a serious confound and it must be stated, not managed.  Part of the reason so many phyla first appear in Stage 3 is that Stage 3 is where the Chengjiang and Sirius Passet deposits sit, and an exceptional deposit is a spike in collecting effort as well as a window on an ecosystem.  Any argument built on raw Stage 3 counts is contaminated by where paleontologists have dug.

And the result that goes against a naive reading of the tables above

Every stage with enough records was cut to the same 500 randomly drawn occurrences, a hundred times over, and the average number of classes and genera recovered was recorded.  This is the standard way to ask what a stage would look like if it had been sampled as hard as every other stage.¹¹

Once sampling is standardised, the ratio of classes to genera bounces between 0.06 and 0.13 with no early peak.  The highest value in the table is Stage 4 and the lowest is the Guzhangian, both well after the phylum-rank first appearances have stopped.  There is no signal here that the early Cambrian was top-heavy compared with the late Cambrian.

So the front-loading in the tables above is a statement about first appearances, not about standing ratios, and this article does not blur the two.  A "first appearance" is a one-time event in the record.  A standing ratio is what a random sample of the sea floor would have looked like at a given moment.  The first is front-loaded; the second is not.  Anyone quoting the first as though it were the second is overstating the case, and that includes this site's own tradition.

The confound this article cannot dismiss: the Sinsk Event

The headline result above is that phylum-rank first appearances stop at the top of Cambrian Stage 3, about 514.5 Ma on the scale the database uses.  There is a named global extinction event in the very next stage, Stage 4, and it has a mechanism.

Zhuravlev and Wood worked from 430 skeletal species on the Siberian Platform and split them, on morphology rather than on a molecular tree, into probable stem-group forms and inferred crown-group forms.  The stem forms appear first and diversify through the earliest Cambrian.  The crown forms appear later and radiate in Stage 3.  Then, in Stage 4 at about 513 Ma, the stem-group fauna is destroyed and never recovers, while the crown groups are barely touched.  Their attribution is specific: the extinction "coincides with the well-known Sinsk Event, an episode of widespread shallow marine anoxia" recorded on the Siberian Platform and elsewhere, which also takes out the archaeocyath sponges.⁵

An extinction can end a wave of first appearances by removing the fauna that was producing them.  If the animals that were still being assigned to new high-rank groups died in a global anoxic event, the count would stop for a reason that has nothing to do with the exhaustion of evolutionary novelty.  Whether "first appearances stop" and "the stem-group fauna is destroyed" are the same event has not been tested here, and it would need a lineage-by-lineage check that nobody appears to have run.

Two things are worth adding rather than hiding.  Zhuravlev and Wood ran a taphonomic control and reported it: all their fossils come from the same acid-dissolution method applied to the same restricted rock types, drawn from many independent studies with no single dominant worker, so the sampling biases are at least shared across their whole dataset.⁵  That is a methodological strength, and it makes their result harder to wave away, in either direction.

And their two-phase reading is itself a challenge to the popular picture.  On their account the explosion is not one event but two: a terminal-Ediacaran-to-earliest-Cambrian phase producing stem groups, and a Stage 3 phase producing crown groups, one above the other in the rock and divided by an extinction.  A single instantaneous appearance of finished body plans is not what their data show.

"All the animal phyla appear at once" is false, and the database says so plainly

This is the claim in the popular creationist literature, and it does not survive.

Ten phyla in the database have their oldest occurrence after the Cambrian:¹¹

And seven living phyla have no fossil record at all: Cycliophora, Gnathostomulida, Micrognathozoa, Gastrotricha, Xenoturbellida, Orthonecta and Rhombozoa.  Zhang and Shu state the same fact from the other side of the argument.⁸

Notice which phyla these are.  Every one of them is soft-bodied, and most are small or microscopic.  That is exactly the pattern the preservation explanation predicts, and this article concedes it rather than dodging it.  Erwin and colleagues put it in one sentence that goes against this site's position: the phyla and classes added to the record after the early Paleozoic "are largely artifactual, as they represent occurrences of taxa with little or no preservation potential."⁴

What replaces the dead claim is narrower and still striking.  Of the animal phyla that have any fossil record at all, roughly half make their first appearance inside the Cambrian, and every one of those that appears in the Cambrian has appeared by the top of Stage 3.

Four caveats a hostile reviewer will raise, stated first

The database's phylum list is not the standard one.  Its accepted phyla for Animalia include entries that no textbook would list: Problematica, an unassigned catch-all, the superphylum Lophotrochozoa, the Ediacaran groups Petalonamae and Proarticulata whose affinities are disputed, and several redundant pairs such as Silicea and Porifera or Priapulida and Scalidophora.  The standard modern count is about 33 to 35 phyla.  The pattern in the tables is what should be quoted; the raw phylum list should not.

Rank is not a natural category.  A phylum's first appearance is, by construction, the first fossil somebody chose to assign to it.  Some front-loading at high rank is guaranteed by the way taxonomy works, and the argument's opponents have said so in print.  The Ordovician's order-and-family output is a partial check against that, and it is a check, not a proof.

Sampling is uneven between stages, by a factor of twenty-seven in well-dated collections, and the table above prints it.

Standardising the sampling removes the early-Cambrian top-heaviness in the class-to-genus ratio altogether, and the table above prints that too.

Part 4: How long, on the mainstream's own clock, and the number this article's own side must stop using

Verdict: in the order of the rock, the core event runs from the base of the Cambrian through Stage 3, and wider readings start lower, in the upper Ediacaran.  On the mainstream timescale the core event takes twenty to thirty million years, with forty as the outside figure.  "Five to ten million years" misstates that timescale and is dead.

This article does not argue from durations, and every figure in this part is the mainstream's.  The figures matter for one reason: creationist and intelligent-design writers quote a duration as though it were the mainstream's, and that claim can be checked against what the mainstream actually says.

There is no single published duration for the Cambrian explosion, and an honest article says so.  What exists is a set of figures that differ because their authors start and stop the clock at different events.  Each of the following was read out of the paper itself.

Marshall states the honest range himself in a single sentence, and it is the fairest summary in the literature for a general reader.  He puts "explosion" in quotation marks because the radiation "still extended over some 20 million years of the earliest Cambrian, or longer if you add in the last 30 million years of the Ediacaran and the entire 55 million year duration of the Cambrian."¹⁶

This article's own result is an interval of rock, not a number of years.  Part 3's strict-containment analysis puts the last phylum-rank first appearance at the top of Stage 3, the third stage of the Cambrian.  The chart the Paleobiology Database follows assigns that interval 24.3 million years, close to Erwin's figure.  That match shows only that the database and Erwin's group use much the same timescale.  It is not independent evidence of a duration, and nothing in this article rests on it.

Zhang and Shu give the widest defensible figure and give it as a maximum, which is the honest way to quote it.  They put the oldest body plans in the record at about 560 Ma and a metazoan-dominated ecosystem in place by about 521 Ma, and conclude that "the maximal duration of the Cambrian Explosion is about 40 Ma."  Their own gloss on that number is the one to borrow: "Forty million years is rapid in the larger scheme of Earth's history, just about 1% of life's history."⁸

An article that quotes forty million years and calls it rapid is quoting the mainstream accurately.  An article that quotes five million years is not.

Where "five to ten million years" came from

The short figure runs through popular creationist and intelligent-design writing.  Its best mainstream source is a 1996 paper by the geneticist Susumu Ohno, and this article verified the sentence: the Cambrian explosion "was of an astonishingly short duration, lasting only 6-10 million years."²⁰

Two things travel with that quotation, and neither is optional.

First, the dating is from 1996 and rests on Siberian geochronology that has since been revised.  The interval Ohno was describing is now assigned a longer span, and no current specialist source supports 5 to 10 million years as the duration of the event.

Second, the very next sentence of Ohno's abstract is his naturalistic solution, and this site's own tradition cuts it.  Ohno's point in quoting a short duration was that only about 1% base-sequence change would be expected in ten million years, so the Cambrian animals must have been "endowed with nearly identical genomes, with differential usage of the same set of genes accounting for the extreme diversities of body forms."²⁰  He called it the Cambrian pananimalia genome.  He was proposing a mechanism, not conceding a miracle.

Quoting Ohno's duration without Ohno's proposal is a quote-mine, and it is one this article's own side commits.  It is also a proposal that later work broadly vindicated.  The shared developmental toolkit that Marshall and Erwin both discuss is Ohno's idea with another twenty-five years of data behind it.

The other source of the short figure is Bowring and colleagues' 1993 dating of the Tommotian and Atdabanian stages of Siberia.  This article did not obtain that paper and does not quote it.  What can be said without it is structural: those two stages are a subinterval of the event, not the whole of it.  Using a subinterval's length as the event's length is the same error as timing a marathon from mile 18 to mile 23.

And the dates themselves are contested at the million-year level, which does not matter here

Erwin and colleagues give the base of the Cambrian as 541 ± 0.13 Ma.  Marshall gives 542 to 543 Ma.  The chart the Paleobiology Database follows gives 538.8 Ma.  That is a spread of about four million years on a boundary quoted to a tenth of a million.

None of it is this article's evidence.  The argument runs on order.  Phylum-rank first appearances are concentrated in the lower part of the Cambrian and stop; ichnodiversity and ichnodisparity jump at the Ediacaran–Cambrian boundary; the preservation windows below that boundary hold algae and not animals.  Every one of those statements survives unchanged if every absolute date on the chart moves.

Part 5: The preservation objection, at full strength

Verdict: the explosion survives this objection as a real event.  Its measured magnitude does not survive as a measured quantity.

This is the explanation that could dissolve the whole argument, so it gets the longest hearing here, stated in its authors' own terms before any answer is offered.

The claim

Animals become fossils when they have hard parts.  If a large number of lineages acquired mineralised skeletons over a short interval, a large number of lineages would appear in the record over a short interval, whether or not anything happened to them evolutionarily.  On this reading the Cambrian explosion is a skeletonisation event recorded as a diversification event.

Mángano and Buatois state it as one of the two historical positions in the field: the view that the Cambrian explosion is an artefact of an increase in the preservability of body fossils, tied to the acquisition of mineralised skeletons and probably to major changes in ocean chemistry.¹

The factual core of the objection, and it is strong

Zhang and Shu tabulate biomineralisation across the transition as a three-phase process, and the phases are not simultaneous.⁸

Now compare that with Part 3.  The order in which phyla first appear in this article's own analysis is molluscs and other lophotrochozoans early, arthropods and brachiopods in Stage 2, echinoderms in Stage 3.  That is the same order in which their skeletons appear, phase by phase.

That correlation is exactly what the preservation explanation predicts, and this article prints it as a genuine problem rather than a coincidence.  Anyone who takes Part 3's tables as a direct read-out of evolutionary events has to explain why they line up so neatly with the sequence of skeleton acquisition.

The diversity numbers move with the skeletons too.  In South China, small shelly fossil genera go from 11 in the Fortunian to 169 in Stage 2, and Stage 3 reaches 431 or more, largely through the Chengjiang deposit, from which more than 280 species have been described.⁸

The second half of the objection: the soft-body window is real, and it is widest in the early Paleozoic

Gaines and colleagues worked out the mechanism of Burgess Shale-type preservation from sedimentology and sulfur isotopes across seven principal deposits, Chengjiang and the Burgess Shale among them.  What they found is a chain of conditions, every one of them a property of the early Paleozoic ocean rather than of the animals.²¹

1. Low sulfate in the global ocean and low-oxygen bottom water at the sites of deposition, which starved sediment microbes of the oxidants they use to destroy soft tissue.
2. Rapid burial in ultra-fine claystone, every particle under 25 micrometres.
3. Carbonate cement laid down at the seafloor as a cap over the bed, forming a barrier that kept oxidants out. This feature is shared by every deposit they examined, and they attribute it to the unusually high alkalinity of Cambrian seawater.
4. Absence of burrowing animals, which the low-oxygen bottom water enforced, and which greatly improved the cement barrier.

Their conclusion is the sentence that matters: this style of preservation "was promoted by unique aspects of early Paleozoic seawater chemistry."²¹

This is a serious argument and it is the best one on the table.  It says the window through which soft-bodied Cambrian animals are visible opened because of a particular ocean chemistry and closed when that chemistry changed, and one of the things that changed it was animals themselves, since burrowing both destroys the cement barrier and raises seawater sulfate.  On this account, the apparent burst of strange soft-bodied forms in the early Cambrian and their apparent disappearance afterwards are both effects of preservation.

And it bears directly on this article's own headline number.  Part 3 found that phylum-rank first appearances stop at the top of Stage 3.  Stage 3 is where Chengjiang sits.  The Burgess Shale-type window is closing over roughly the same interval.  A hostile reviewer will say that "the count stops dead" is the sound of a preservation window shutting, and that objection belongs on the same page as the result.  It is on this one.

The answers, in order of strength

1.  Burrows need no window, and they give the same answer.  This is Part 1's material and it is the strongest reply available.  Ichnodiversity goes from 7 to 42 and ichnodisparity from 5 to 22 across the Ediacaran–Fortunian boundary, in a record two ichnologists describe as far more continuous than the body-fossil record.¹¹  A burrow requires no skeleton, no anoxic bottom water and no seafloor cement.  Whatever the taphonomic window does to body fossils, it cannot reach a metre-deep vertical shaft in ordinary sediment.

2.  The Ediacaran windows exist and contain no animals.  Part 2's material.  The Miaohe biota is a latest-Ediacaran Burgess Shale-type deposit and it is richly fossiliferous, with algae.¹³  The Doushantuo phosphate preserves cellular detail and preserves no arthropods.  Erwin and colleagues make the same point for skeletons: Ediacaran phosphate deposits are common and contain no small shelly fossils.⁴  If preservation mode were the only thing standing between the record and a Precambrian animal fauna, these deposits should hold one.  And the authors of the standard monograph add that ordinary preservation may have moved the wrong way for the objection:  once burrowing animals replaced the microbial mats, "the quality of fossil preservation in some settings may actually have declined from the Ediacaran into the Cambrian."²²  Part 9 gives the passage in full, with the authors' own view of a long earlier history beside it.

3.  Skeletonisation is itself the thing to be explained.  Saying that the phyla appear when they acquire skeletons restates the observation unless one can say why a dozen unrelated lineages acquired skeletons within the same few stages at the base of the Cambrian, having gone without them through all the fossil-bearing rock below.  Zhang and Shu treat episodic biomineralisation as a component of the explosion, not as an alternative to it.⁸

4.  The window explanation predicts the wrong thing about the Ordovician.  Part 3's rank table shows the Ordovician producing more first-appearing orders and families than the Cambrian, while producing one phylum and seventeen classes.  The Ordovician has no Burgess Shale-type window.  If the Cambrian's high-rank first appearances were a window artefact, the Ordovician's low-rank first appearances should have been suppressed by the same mechanism.  They were not.

What the answers do not reach, and this has to be said

The four replies establish that the Cambrian explosion is not wholly a preservation artefact.  They do not establish how much of it is.

Nobody has partitioned the signal.  There is no published estimate of the form "X% of the apparent Cambrian rise in first appearances is taphonomic," and this article could not construct one, because it would require a preservation model for each lineage and no such model exists.  Part 3's sampling table shows well-dated collections varying by a factor of twenty-seven between stages.  Part 3's standardisation table shows that levelling the sampling removes the early-Cambrian top-heaviness in the class-to-genus ratio entirely.

So: the event survives; the size of the event does not.  This article claims the first and declines the second.  Anyone who tells a reader that a precise percentage of animal body plans appeared in a precise number of million years is claiming a measurement that has not been made.

Part 6: Disparity, stem groups, and the results that go against this article

Verdict: the popular version of the disparity argument is dead, the stem-group objection is the strongest one this article faces, and the best recent measurement says the gaps were produced by extinction.

What this article does not claim, and why

Nearly every popular version of the Cambrian argument, creationist and secular alike, descends from Stephen Jay Gould's Wonderful Life (1989).  Gould made several claims there about the range of anatomical design in the Burgess Shale.

Two of them were measured in 1994 and did not survive.  Wills, Briggs and Fortey compared Cambrian arthropod disparity against modern arthropod disparity using four phenetic and two cladistic methods, and reported that none of those methods finds markedly greater disparity in the Cambrian sample.²³  The series flagship works through that measurement, and through the larger 2013 test that followed it, in detail.⁹

This article defers to the flagship and does not restate the popular version of Gould's claim anywhere.  No argument here rests on it.  The version of it that circulates in creationist and popular writing was tested by the specialists who worked on the same fossils, and it did not survive the test.  Repeating it after 1994 is not something this site is willing to do.

Two things in Gould's book do survive, and both are under-used.

Gould's own control experiment, which holds

Gould anticipated the obvious objection to his own case, that the Burgess looks strange only because soft-bodied preservation is rare, and tested it against the Solnhofen Limestone, the other famous exceptionally preserving deposit, the one that yielded Archaeopteryx.  His finding: "the Solnhofen contains nothing, not a single animal, falling outside well-known and well-documented taxonomic groups."³

That is a genuine matched comparison and it is a good argument.  Two deposits, comparable preservation quality, radically different taxonomic outcomes.  It does not show that the explosion was instantaneous.  It does show that "exceptional preservation reveals unclassifiable animals" is not a general law of exceptional deposits.

The limitation, stated alongside it: the Solnhofen is Jurassic.  It samples a world in which the surviving groups had already been sorted out by every extinction between the Cambrian and the Jurassic, which is exactly what a decimation model predicts.  So the comparison is consistent with Gould's own explanation as well as with this article's.

Gould's own measurement, which cuts against the apologetic use of Gould

This deserves far more attention than it gets.  With Gilinsky and German, Gould measured 708 spindle diagrams of genera within families across the whole marine invertebrate record, scoring each by its centre of gravity, a value below 0.5 meaning the group peaked in the first half of its history.  Lineages arising in the Cambrian or Ordovician came out bottom-heavy.  Lineages arising later could not be distinguished from symmetrical.  He also found the same bottom-heaviness in Paleocene mammals, and concluded that the pattern is not an oddity of Cambrian invertebrate life but a general feature of evolutionary diversification.³

Gould's own data say the Cambrian pattern is not unique.  Anyone citing Gould for "the Cambrian is unlike anything else in the record" is citing against Gould's own measurement.  The larger 2013 test reached the same place from a much bigger sample.⁹

The Burgess Shale, described by the man who described it

Conway Morris's 1979 review is worth reading before any secondary account.  Walcott shipped more than 40,000 specimens to the Smithsonian from the 1910–13 and 1917 field seasons.  The fauna is about 119 genera and 140 species, of which arthropods are 37% of genera.  Of the non-trilobite arthropods, the majority of genera could not be placed in any higher taxon.²⁴

The sentence worth quoting is the ecological one.  Conway Morris writes that the fauna "apparently lived on the edge of the open ocean close to the paleo-equator; it seems neither atypical nor aberrant."²⁴  That is a specialist saying, in the primary literature and a decade before Gould's book, that the Burgess assemblage was an ordinary community in an ordinary setting.  It supports Gould's Solnhofen argument from an earlier and independent source.

And a caution for this article's own side.  "Cannot be placed in any higher taxon" meant, in 1979, "cannot be placed in a Linnaean class," at a time when stem-group methods were not in general use.  Most of those genera are now classified as stem-group arthropods.  The 1979 count is a statement about the taxonomy of 1979.  It is not a measurement that stands today, and it should not be quoted as one.

The stem-group objection, at full strength

This is the objection that does the most damage to any argument phrased in phyla, and this article states it without softening.

A living phylum is a crown group: the clade descended from the last common ancestor of its living members.  Everything on the branch below that point is the stem group: animals with some of the phylum's diagnostic characters and not others.  Those are precisely the intermediates a gap argument says are missing, and the Cambrian is full of candidates.

Zhang and Shu put the mainstream position in one sentence: "The high number of stem groups in the early history of animals is a major reason for morphological gaps across phyla that we see today."⁸  They give the worked example.  Panarthropoda contains three living phylum-rank groups: tardigrades, velvet worms and arthropods.  Cambrian forms such as Anomalocaris and Opabinia sit, on current phylogenies, between the three, and each of them was formerly treated as a body plan of its own.⁸

Verdict, and it goes against this article's side.  The reclassification of the Cambrian oddities from "separate phyla" to "stem groups between living phyla" is not a bookkeeping trick.  In the cases Zhang and Shu name, the animals really are morphologically intermediate between living phyla.  An argument of the form "N phylum-level body plans appear at once with nothing bridging them" cannot survive contact with the literature on Anomalocaris and its relatives.

Which means the claim "phyla appear suddenly" is in part a function of assignment practice rather than purely of the rocks.  The rock does not contain phyla.  It contains animals.  Whether a given Cambrian animal counts as a new phylum, a stem group, or an aberrant member of a known group is a judgement a taxonomist makes, and the judgement has moved steadily in one direction for forty years.  What survives is the narrower result in Part 3: that first appearances, as recorded and as filed, are strongly front-loaded by rank, and even that is a statement about a filing system as much as about a fauna.

Whether each named stem group is morphologically intermediate rather than merely phylogenetically basal is an empirical question that has to be settled case by case.  The named transitional-sequence articles in this series exist for exactly that, and Anomalocaris and the lobopodians are the test cases here.

The measured result that goes furthest against this article

Deline and colleagues built a morphospace across the whole animal kingdom from a large character set, placed both fossils and reconstructed phylogenetic ancestors into it, and asked whether the clumpiness of animal form is original or produced.  Their answer, in their abstract: the clumpiness of morphospace occupation by living clades "is a consequence of the extinction of phylogenetic intermediates, indicating that the original distribution of morphologies was more homogeneous."²

They also report that arthropods, chordates, annelids, echinoderms and molluscs have continued to explore and expand the limits of morphospace throughout the Phanerozoic, well beyond the envelope occupied in the Cambrian.²

This is the strongest published counter to the whole disparity-first framing and this article prints it in full.  It is a direct measurement.  It postdates almost everything in the apologetics literature on the subject.  And it says the two things a Cambrian-explosion argument least wants to hear: the gaps were made by extinction rather than present from the start, and the total range of animal form has grown since the Cambrian rather than shrunk.

Zhang and Shu, who argue that the explosion is a real and distinct event, cite it approvingly and reach a compatible conclusion of their own: the envelope of animal form was already broad in the early Cambrian, did not reach its maximum there, and has not shrunk significantly through extinction since.⁸

The honest statement, in one sentence.  Early high disparity is a real and majority tendency, it is not unique to the Cambrian, it is not exceptionless, the total envelope of animal form has expanded since the Cambrian, and the gaps between living phyla were largely produced by the extinction of intermediates rather than present at the start.

What is left of the disparity argument

Two things, and they are narrower than the popular version.

First, the front-loading of first appearances by rank, which is Part 3's computed result and does not depend on any morphospace at all.  Half of fossilised animal phyla and half of fossilised animal classes first appear inside the Cambrian; 13% of orders and 6.6% of families do.

Second, the order in which the categories fill up inside the Cambrian.  By the top of Stage 3, every one of the Cambrian's phylum-rank first appearances and three-quarters of its class-rank first appearances have happened, against fewer than half of its genus-rank ones.  Whatever one thinks of ranks, the sequence runs from coarse to fine.  That is the reverse of the order a slow accumulation of small differences would produce.

And one concession offered without being asked.  Part 3's standardisation table shows no early peak in the class-to-genus ratio inside the Cambrian.  This article's own data do not support the claim that the early Cambrian was top-heavy compared with the late Cambrian, and the claim is not made here.

Part 7: What has been proposed to explain it, and how each proposal scores

Verdict: after more than a century and a half, no proposed cause is regarded by specialists as sufficient to explain the origin of the animal body plans, and that is a much weaker statement than it sounds.

The most useful service this article can do a skeptical reader is to lay out what has been proposed and report how well each proposal is thought to work by the people who work on it.

Charles Marshall's 2006 review in the Annual Review of Earth and Planetary Sciences is the standard audit.  He built a table scoring each proposed cause against each thing that needs explaining, colour-coded: green for necessary and sufficient, yellow for necessary but not sufficient, orange for consistent but with importance not established, red for not in accord with the fossil record, grey for assuming the result rather than analysing it.¹⁶

No explanation in Marshall's table scores green for disparity.  That is the single most important fact in this part.

The five things that need explaining

Marshall's own list, and it is worth adopting because it makes the argument tractable.¹⁶

1. The increase in disparity: the origin of the body plans themselves.
2. The increase in diversity: the number of species.
3. The time of onset: why then.
4. The duration: why tens of millions of years rather than one or a hundred.
5. The uniqueness: why it never happened again.

He adds two problems that surface on close reading: where the fossils of the basal bilaterian stem groups are, and what the Ediacarans were.

Environmental explanations

Marshall's summary of the whole environmental class is the sentence to keep: none of them address why an environment permissive of complex animal life should necessarily lead to complex animal life.¹⁶

Developmental explanations

Ecological explanations

The mainstream's own bottom line

Zhang and Shu, reviewing the whole field in 2021, report that the proposed intrinsic and extrinsic causes are prerequisites rather than triggers, that no ecological driver is unique to the Cambrian and so none explains why the event happened when it did, and that the explosion is a many-sided event in natural history.  Their conclusion: "No simple, single cause can explain the entire phenomenon."⁸

How this article uses that, and how it must not

It may say, accurately and with citations, that after more than 165 years no proposed cause is regarded by specialists as sufficient to explain the origin of animal body plans, and that the standard review of the subject scores every candidate as necessary-but-not-sufficient at best.

It may not say that specialists therefore doubt that it happened by natural means.  Marshall, Erwin, Zhang and Shu all propose combinations of natural causes and all expect the answer to be found.  An argument from the absence of a settled mechanism is an argument from ignorance, and this site says so about arguments from ignorance in other people's hands.

The honest form of the claim is narrower and still worth making: the explanatory situation is much weaker than popular science writing implies, and the specialists say so in print.  A reader who has been told that the Cambrian explosion is well understood has been told something the standard reviews do not support.

The young-earth reading of the same rocks

Verdict:  it fits the outline of the Cambrian base as well as any trigger in the tables above, and it is weakest on the order inside the transition, which it has not yet been tested against.

Young-earth geologists do not look for a cause of the Cambrian explosion.  They read it as the place where the Flood first buried sea-floor life on a large scale.  Kurt Wise puts it plainly:  "Most creationists are not so much bothered by the Cambrian Explosion as they expect such a paleontological discontinuity with the initiation of Flood sedimentation."²⁷  Timothy Clarey of the Institute for Creation Research and Davis Werner tie it to the same erosion surface Peters and Gaines describe:  "The Sauk megasequence likely was only part of a violent beginning to the Flood (during the first 40 Days), creating the Great Unconformity at its base and encasing prolific numbers of hard-shelled marine fossils, defining the Cambrian Explosion."²⁸

They do not agree on where the Flood begins.  Andrew Snelling, with Wise and Steven Austin, puts the start lower, at beds of broken rock in the Grand Canyon and the Mojave Desert that lie below the Ediacaran fossils.  Snelling notes that this is "not all that far below the Precambrian/Cambrian boundary, the traditional placement of the beginning of the Flood."²⁹  Wise builds his own model on the same boundary.²⁷  On that placement the Ediacaran organisms are early Flood burials, not fossils of a world before it.

Wise's model is the one that speaks to this article's evidence.  He reads the Mojave sequence as zones of one wide pre-Flood continental shelf, buried in turn as the water rose:  "from the stromatolite-dominated hydrothermal biome, to one where the Ediacaran organisms lived, to one where the organisms lived which generated the small, shelly fossils, and finally to the trilobite-brachiopod-dominated waters farther to the east."²⁷  He also explains why the Ediacaran biota vanishes rather than grading into the Cambrian fauna:  its extinction "is explained by the failure to regenerate the same protective regime in post-Flood times."²⁷

What the model accounts for, measured against this article's findings.

1. An abrupt, marine, world-wide appearance of shelled animals on top of a continent-wide erosion surface. The mainstream describes the same thing: a surface cut across most continents, then "the first continental-scale marine transgression of the Phanerozoic."²⁵
2. An Ediacaran biota that dies out instead of supplying ancestors (Part 2).
3. Uniqueness, the fifth item on Marshall's list. A single event happens once. In the tables above only niche saturation and developmental entrenchment address uniqueness, and each carries the problem the tables give it.
4. Rising water spreading inland. Peters and Gaines report that the Sauk rocks are older at the continental margins and younger in the interiors.²⁵ Clarey reads the same pattern as "the Floodwaters rose progressively as described in Genesis chapter 7."²⁸ A slow sea-level rise predicts the same pattern, so this fact fits both models and decides nothing between them.

Where it is weak, and the weaknesses come from this article's own data.

1. The order inside the transition is the same everywhere. Wise's zones have to be buried in the same order on every continent: Ediacaran organisms, then the small shelly fossils in the phases of Part 5, then trilobites, then echinoderms. The stage order this article counts is the one Wise and Richardson tested against the same database and called real.¹⁰ Zonation explains which broad habitat is buried first. It does not yet explain why one sequence of faunas inside a single shallow-marine setting recurs world-wide. Clarey and Jeffrey Tomkins concede the general form of this in print: "Questions about why distinctly different fossil assemblages are found in the Sauk, Tippecanoe and Kaskaskia Megasequences remain, even though all are dominantly marine."³⁰ Wise's own link between zone and rock type is offered with a "perhaps" and question marks: sand for the Ediacaran zone, carbonate for the small shellies, mud for the trilobites.²⁷
2. The burrows record time at the surface. Part 1's trace record stacks Ediacaran mat surfaces under Cambrian surfaces with new burrow designs. Snelling's own rule is that a burrowed layer "would represent the passing of at least a few hours" before the next layer arrived.²⁹ In the model's favour, early Cambrian burrowing is sparse and shallow. On the ichnofabric index, where 1 means undisturbed bedding and higher numbers mean more burrowing, lower Cambrian strata "only rarely exceed ii 3 and are most commonly characterized by ii 1 or 2,"¹⁵ and in the trilobite-bearing lower Cambrian carbonates of the western United States burrows averaged 2 to 3 centimetres deep.¹⁴ That is the least demanding part of the column for a short timescale. Across the whole Cambrian it is heavier: in 810 metres of inner-shelf Cambrian carbonate there, the commonest index values were 3, 4 and 5, and burrowing increases upward through the period.¹⁴ Bioturbation and Trace Fossils prices burrowed beds against Flood timetables. No bed-by-bed count of burrowed surfaces for a section spanning the Ediacaran–Cambrian boundary was found, so that count is not priced here.
3. Wise names the deeper problem himself. The rocks below the boundary hold bacteria, protists and algae and no definite animals, and he writes that this "does create a challenge for creationist theories of earth history."²⁷ His hydrothermal biome is his answer to it, and it is a proposal about one region that he extends world-wide with a "may."

How strong it is.  As an outline, the young-earth reading fits the Cambrian base at least as naturally as the mainstream triggers:  it predicts an abrupt marine burial on a world-wide erosion surface, and it does not need a cause for body plans, because on its view they were made, not developed.  A skeptic will reply that the model was fitted to the rocks after the fact, and that "they were created" explains any pattern equally well.  A creationist can answer that every trigger in Marshall's table was also proposed after the fact and none of them explains disparity.  Both replies are fair.  What would make the young-earth reading a tested model is a zone-by-zone prediction of the order of first appearances, checked continent by continent against the database.  That has not been published, and until it is, the reading is a plausible outline with untested details.

Part 8: The claims on this article's own side, audited

Each claim below is stated as creationists state it, then checked.  Three of the six do not survive.

"Pikaia is a phylum-level intermediate": dead, and misattributed

This is the most important verdict in the article, and it costs this site more than anything else on the page.

Kurt Wise's 1995 paper is the strongest concession young-earth creationism has produced on transitional fossils.  Wise took his doctorate in paleontology at Harvard under Gould.  He wrote in a creationist journal that three of Darwin's four expectations about intermediate fossils have been confirmed, that creationists have to accept it, and that traditional creation theory predicted none of it.⁶  A project-wide audit for this series found something else about that paper: Pikaia is its only phylum-level example.  It therefore carries the whole weight of the concession at the one rank where a creationist has anything to lose.

Wise's sentence lists his examples of confirmed stratomorphic intermediates, and the relevant parenthesis reads "Pikaia (between echinoderms and chordates)."⁶  His reference for it is Gould's Wonderful Life, pages 321 to 323, the section Gould titles "An Epilogue on Pikaia."

Those pages say the opposite.  Gould's conclusion there is: "Pikaia is not an annelid worm.  It is a chordate, a member of our own phylum—in fact, the first recorded member of our immediate ancestry."³

Gould places Pikaia inside a phylum.  Wise describes it as sitting between two.  Those are different claims, and the second is not on the pages cited.  Echinoderms and chordates are both deuterostomes, so an animal "between" them would be a basal deuterostome lacking chordate characters, and the characters that caused Pikaia to be reclassified in the first place were the notochord and the segmented muscle blocks, which are chordate characters.

Verdict: misattributed.  The parenthesis is Wise's own gloss.  It is not supported by the pages he cites, and it should not be repeated on this site.

And Pikaia has moved twice since, both times in the same direction.

In 2012 Conway Morris and Caron redescribed the animal from 114 specimens and placed it as a stem-group chordate, the most stemward chordate then known.  This article read the abstract and secondary reports of that paper; the full text was not obtained.³¹

In 2024 Mussini and colleagues reinterpreted the anatomy again.  Their deposited abstract reports evidence for a gut canal and for a dorsal nerve cord, which is a chordate character that had not previously been demonstrated in the fossil; a new anatomical model on which the animal had previously been interpreted upside down; a muscle-block arrangement intermediate between amphioxus and vertebrates; and a phylogeny in which vetulicolians form a grade of the earliest-diverging stem chordates, subtending a more derived grade containing Yunnanozoon and Pikaia.  This article could not obtain the full text of that paper, so nothing is quoted from it and everything above comes from the authors' own deposited abstract.⁷

What that does to the argument, in both directions.

Against Wise's version: Pikaia is not between phyla.  It is inside the chordate stem lineage, which is inside the chordate total group.  The one phylum-rank entry in the best creationist concession on this subject does not do phylum-rank work.

Against this article's side more broadly, and this is the harder half: the 2024 result is worse for creationism than the 1989 one was, not better.  A single odd fossil sitting "between echinoderms and chordates" would be one data point.  What Mussini and colleagues describe is a graded series: vetulicolians, then Yunnanozoon, then Pikaia, then crown chordates, with diagnostic characters appearing in order along it.  If that phylogeny holds, it is a documented stepwise assembly of a phylum-level body plan, which is exactly what a creationist account says the record does not contain.

The consequence for Wise's concession has to be stated too.  With Pikaia removed, the concession has no surviving phylum-level entry.  Its force lies at class level instead: the mammal-like reptiles, the tetrapods, the whales, the Cenozoic mammal sequences.  That is still a substantial concession, and still one that traditional creation theory did not predict.  But nobody should cite Wise for a phylum-rank admission, because after this correction there is not one.

The objections to the 2024 result, and none of them is a refutation.  It rests on a character matrix, and character matrices are where phylogenetic disputes live.  Vetulicolian affinities have been argued over for twenty-five years.  And reorienting a fossil by 180 degrees is the kind of reinterpretation that has been reversed before: Hallucigenia was published upside down and back to front, and both errors were later corrected.  Those are real reasons to hold the 2024 phylogeny provisionally.  They are not reasons to prefer a reading of Pikaia that its own source never supported.

"All the animal phyla appear at once in the Cambrian": dead

Covered in Part 3.  Ten phyla in the database have their oldest occurrence after the Cambrian, running from Sipuncula in the Middle Ordovician to Pogonophora in the late Eocene, and seven living phyla have no fossil record at all.  What replaces the claim is narrower: of the animal phyla that have any fossil record, roughly half first appear inside the Cambrian, and every one of those has appeared by the top of Stage 3.

"No new phyla have appeared since the Cambrian": survives, with a definitional caveat

Marshall lists this among the five things needing explanation, and states it sharply: "Even after the largest mass extinction of the Phanerozoic, the end-Permian mass extinction, no new phyla appeared.  Why?"¹⁶  Zhang and Shu treat the same fact as the strongest reason to regard the explosion as a real and distinct event.

The caveat is the one from Part 3.  Phyla first appear in the record long after the Cambrian: nematodes in the Devonian, rotifers in the Eocene.  Nobody thinks those phyla originated then.  They are soft-bodied groups whose fossil record begins whenever a preservation accident happens.  So the claim is about origination, not about first appearance, and this article does not slide between the two.  Erwin and colleagues say exactly this, and it is a concession that has to be printed: the phyla and classes added after the early Paleozoic "are largely artifactual, as they represent occurrences of taxa with little or no preservation potential."⁴

"The Ediacaran organisms are not animals": dead

Covered in Part 2.  Kimberella is a bilaterian-grade animal on body-plus-trace evidence, and Erwin and colleagues assign it to the crown Metazoa.  Dickinsonia's biomarkers run 93% cholesteroids against an algal background of about 11%, which kills the lichen reading outright.

What survives is much weaker and still worth saying.  Seven of the nine likely Ediacaran clades cannot be placed in the animal kingdom with confidence.  The Ediacaran biota goes extinct rather than grading into the Cambrian fauna.  And no Ediacaran organism has been identified as the ancestor of any Cambrian phylum.

"Paleontologists admit the explosion is real and unexplained": survives, and it is the strongest citable claim here

Verified from four independent mainstream sources.  Mángano and Buatois: trace-fossil information "strongly supports the Cambrian explosion."¹  Gaines and colleagues: the taphonomic window is real, and it does not manufacture the trace record.²¹  Marshall: no proposed cause scores as sufficient for disparity.¹⁶  Zhang and Shu: "No simple, single cause can explain the entire phenomenon."⁸

This article may state that and cite it.  It may not convert it into a positive argument for design without saying that it is doing so, and without meeting the god-of-the-gaps objection head-on, which is precisely the objection Marshall levelled at Meyer in Science, and which this site's own rules make it take seriously.  The objections section below answers it directly rather than leaving a reader to raise it.

Three claims in this project's own notes that should be cut

The "Cambrian rabbit" anecdote.  This project's research notes carry a report that a named paleontologist said in a lecture that no fossil find could falsify Darwin's theory.  The source is a creationist writer describing a talk he attended.  A second-hand report of an unrecorded spoken remark, relayed by a partisan, is not a citation.  Unsourceable.  Cut it.

A one-word alteration in a Gould quotation.  This project's notes render Gould's summary sentence as "The sweep of anatomical diversity reached a maximum right after the initial diversification of multicellular animals."  The text reads "anatomical variety."³  The substitution is small, and it is exactly the wrong word: that chapter's whole argument turns on the distinction between diversity, meaning number of species, and disparity, meaning difference in body plan, and Gould avoids "diversity" there for that reason.  The check was made against an optically scanned text, so the wording should be confirmed against a printed copy before this correction is quoted elsewhere.

"Undispersed."  This project's notes attribute the word "undispersed" to Erwin and Valentine.  The term in the book is underdispersed, and the notes took it from a creationist website rather than from the book.  The flagship covers the correction.⁹

Part 9: Darwin's Doubt, audited, and its critics audited too

Stephen Meyer's Darwin's Doubt is the most prominent intelligent-design book on this subject, and any article on the Cambrian will be read against it.  This part audits the book and the three published criticisms of it that get quoted most.  It corrects all three critics before it faults the book, because a reader who watches this site correct hostile critics of a creationist book has reason to believe it when it says the book has a real problem.

First, the edition

The copy used here is the 2014 paperback, whose copyright page reads "Copyright © 2013 by Stephen C. Meyer.  Epilogue © 2014 by Stephen C. Meyer."³²  That paperback carries a long epilogue answering critics of the first edition, and the epilogue does not exist in the hardcover.  Any claim of the form "Meyer never mentions X" has to specify an edition, and the counts below distinguish first-edition text from 2014 additions by position in the file.

The word counts, which are the checkable part of the critics' charge

Cloudina is the best-known skeletonised organism from below the Cambrian boundary, the animal that shows biomineralisation beginning in the Ediacaran.  The small shelly fossils are the assemblage of tiny plates, tubes and spines that spans the earliest Cambrian, below the first trilobites.  Both are directly relevant to a book arguing that skeletonised animals appear abruptly.  Neither is in the main text.

The three critics, one at a time

Nick Matzke, 2013: largely correct, and correct in the specifics.  He wrote that Cloudina "gets no mention at all in Meyer's book," that "Tommotian" does not appear in the book, and that the small shelly fauna gets one mention, buried in an endnote to chapter 4.³³  Checked against the text: Cloudina is absent from the book entirely, bibliography included.  "Tommotian" was absent from the first edition and appears exactly once in the 2014 epilogue.  The small shelly fauna's only first-edition appearance is a chapter-4 endnote.  Matzke's note number is off by twelve, endnote 39 rather than 27, and that is the only error in the charge.

Charles Marshall in Science, 2013: right in substance, wrong in one word.  He wrote that "Meyer completely omits mention of the Early Cambrian small shelly fossils."³⁴  There is an endnote, so "completely" is not literally true.  The substance survives: the small shellies appear nowhere in the main text, nowhere in the figures, and nowhere in the passages where the duration of the explosion is argued.

Donald Prothero, 2013: the same shape of error, and less defensible.  He wrote that Meyer "completely ignores the existence of the first two stages of the Cambrian (nowhere are they even mentioned in the book, or the index)."³⁵  The Manykaian, the first of those two stages, is named, dated and described accurately in endnote 39.  Wrong as written.  Prothero's underlying charge is half right for the wrong reason: the Tommotian, the second of the two, genuinely is absent from the first edition.

Meyer's central number, and why it does not survive

The book's quantitative claim is stated twice.  In the main text, Meyer writes that the main pulse of Cambrian morphological innovation occupied no more than six million years, and that Erwin and colleagues "similarly show that thirteen new phyla appear in a roughly 6-million-year window."  In the 2014 epilogue he restates it: "Darwin's Doubt focuses on the crucial Tommotian and Atdabanian stages of the Cambrian explosion—where 13–16 new animal phyla arose within a 5–6 million year window."³²

Three problems, in ascending order of seriousness.

1.  The timescale has moved, and by a factor of about two and a half.  Meyer's figures descend from 1993 geochronology on the Siberian Platform.  The Tommotian is now correlated with Cambrian Stage 2 and the Atdabanian with Cambrian Stage 3.  On the chart the Paleobiology Database follows, those stages run 529 to 521 Ma and 521 to 514.5 Ma, about 14.5 million years together, not 5 to 6.  Correlating Siberian stages to the global scale is imprecise and this article says so, but the direction of the revision is not in doubt.  The interval Meyer names is now assigned roughly two and a half times the length he gives it.  Every figure here is the mainstream's; the point is that Meyer misreports the mainstream timescale he is citing, not that this article vouches for it.

2.  Erwin's own reading of Erwin's own figure is 25 million years, not 6.  Meyer's endnote points to Erwin and colleagues' 2011 paper in Science.  That paper's own summary of the same data is "a rapid (~25-million-year) evolutionary divergence of most animal clades near the base of the Cambrian."⁴  Counting thirteen first appearances inside a six-million-year slice of Erwin's figure is not fabrication, since the data are Erwin's, but presenting it as what Erwin shows, while omitting Erwin's own summary of the same figure, is selective.  Verdict: context-stripped.

3.  The figure conflicts with every duration in the current literature: Marshall's 20 and 65, Erwin's 25, Daley's 30, Zhuravlev and Wood's 29, and Zhang and Shu's 40.  A Berean Archive article must not use it, and this one says plainly that the most-cited design book on the subject rests on a number no current source supports.

One more quotation problem, in Meyer's direction

Meyer quotes Marshall's 2006 review in a way that splices Marshall's main text with a footnote to Marshall's summary table, twenty pages apart.³²  The endnote discloses it only by carrying two page numbers.  Marshall complained in print that a quotation of his review was assembled from two very different parts of it.³⁴  Checked, and the complaint is justified.

Meyer's best material, checked, and it holds

Meyer's strongest reply to the preservation objection is a passage he quotes from Erwin and Valentine's 2013 monograph, the standard book-length account of the subject.³²  The passage is real, it is on the page Meyer cites, and he quotes its substance faithfully.  Erwin and Valentine write that "a revolutionary change in the sedimentary environment—from microbially stabilized sediments during the Ediacaran to biologically churned sediments as larger, more active animals appeared—occurred during the early Cambrian.  Thus, the quality of fossil preservation in some settings may actually have declined from the Ediacaran into the Cambrian, the opposite of what has sometimes been claimed, yet we find a rich and widespread explosion of fauna."²²

Two small faults are Meyer's.  His quotation swaps two words ("may have actually declined from the Ediacaran to the Cambrian").  And his lead-in says that "many" late Precambrian settings favoured preservation, where the authors wrote "in some settings."  The substance is unchanged by either.

Why it matters.  It is a mainstream statement, by the two authors of the standard monograph, that the preservation variable moves the wrong way across the boundary for the preservation explanation.  It sits in a paragraph that lists the reasons Erwin and Valentine accept the explosion as real, which include acritarchs and trace fossils, each "subject to unique preservational requirements" and so "independent metrics of diversity change."²²

What it does not mean, and it must travel with the quotation.  Erwin and Valentine are not doubting animal evolution or a long prior history.  Two pages earlier they write that "A long history of metazoan evolution did precede the Cambrian, perhaps by 200 million years or more."²²  Their point is that the Cambrian fossil burst is not produced by a sudden improvement in preservation.  Their claim is limited to "some settings," and it does not touch the Burgess Shale-type window, which Part 5 treats separately.

Marshall's scientific objection, Meyer's reply, and who is right

Marshall's charge is that Meyer's case rests on new body plans requiring large amounts of new genetic information, including many new protein folds, whereas the current understanding of morphogenesis is that new phyla emerged largely through the rewiring of gene regulatory networks built from genes that already existed.  Meyer dismisses network rewiring on the grounds that manipulating today's networks is lethal.  Marshall answers that today's networks have been overlain with half a billion years of subsequent innovation, and that early ones were not so encumbered.³⁴

Meyer's reply, in the 2014 epilogue, is that rewiring presupposes three sources of information it does not explain: the information in the network's genes, the information required to rewire them, and the information in the pre-existing protein-coding genes.

The verdict, and it favours neither cleanly.

Marshall's "not so encumbered" answer is an assertion rather than a measurement, and he cites no evidence for it in the review.  And Marshall's own 2006 review reported the underlying idea as unresolved.  There he listed developmental entrenchment as a very appealing idea about which much remains to be learned, and cited a Silurian trilobite with Cambrian-level variability in segment number as evidence against it.¹⁶  So Marshall's 2013 rebuttal leans on the very hypothesis his own 2006 review reported as unsettled and partly contradicted.  And if entrenchment is weak, then "early networks were more flexible" is weak too.  That is a fair point for this article's side and it is checkable in Marshall's own two publications.

Against that, Meyer's argument depends on protein-fold rarity estimates that are contested elsewhere on this site, and Marshall's core empirical claim is well supported.  Erwin and colleagues state it independently: the morphological differences between sponges and arthropods "cannot be due to the absence of these protein-coding gene families."⁴  The information argument belongs to this site's genetics articles and not to this one.  The dispute is stated here; it is not settled here.

Matzke's strongest point, which this article's side has to answer

Matzke's main charge is not about word counts.  He lays out a graded Ediacaran-to-Cambrian sequence: microscopic algae, sponge-grade fossils, cnidarian-grade fossils, surface tracks, Cloudina and the first drilled predation holes, vertical burrows, small shelly fossils diversifying, and only then the first body fossils assignable to phyla, most of them on stems rather than in crowns.  He says a reader of Darwin's Doubt gets none of it.³³

Checked against the primary literature independently, that sequence is real.  Parts 1, 2 and 5 of this article reconstruct it from Mángano and Buatois, Erwin and colleagues, Daley and colleagues, and Zhang and Shu, without reference to Matzke.  It is graded, it is ordered, and it runs from the Ediacaran up through Cambrian Stage 3.

An article that presents the Cambrian as a single instantaneous appearance of finished body plans is not describing the record, and this one does not.  What survives after Matzke's sequence is granted is Part 3's rank structure and Part 1's jump in ichnodisparity: both real, both measured, and both much narrower than the popular claim.

Matzke's own weakest move is worth recording.  He calls the sub-crown Chengjiang arthropod-like forms transitional forms and asks how the fact can go unmentioned.  Whether a stem-group fossil is morphologically intermediate rather than merely basal on a tree is the empirical question Part 6 flags, and it has to be settled case by case rather than asserted from a phylogeny.

Objections and responses

The objections below come from both directions.  The ones from this article's own side are answered as sharply as the others.

1. "The Cambrian explosion is just when animals evolved hard parts."

This is the strongest objection to the whole article and Part 5 gives it a full hearing.  The short answer is that burrows have no hard parts.  Ichnodiversity goes from 7 to 42 and ichnodisparity from 5 to 22 across the Ediacaran–Cambrian boundary, and maximum burrowing depth from a centimetre to a metre, in a record two ichnologists describe as far more continuous than the body-fossil record.¹¹  A metre-deep burrow in ordinary sediment is not a taphonomic effect.

The second answer is that the latest-Ediacaran preservation windows exist and are not empty.  The Miaohe deposit preserves soft tissue and contains algae; the Doushantuo phosphate preserves cellular detail and contains no arthropods.¹³

And the part of the objection that lands: nobody has partitioned the signal, and this article says so.  How much of the measured magnitude of the explosion is taphonomic is unknown, and Part 5 declines to claim otherwise.

2. "Molecular clocks put the animal phyla deep in the Precambrian, so the fossils are simply late."

This is the largest evolutionary answer to the whole subject, and this article does not settle it.  Molecular clock estimates place the last common ancestor of living animals well before the Cambrian and most phylum-level stem lineages in the Ediacaran or earlier.  That is a separate body of evidence with its own methods, its own calibration problems and its own article on this site, Do Clock Dates Agree with the First Fossils?.  Nothing here should be read as a refutation of it.

Two things can be said from within this article's evidence.  First, the tension is the mainstream's own, and it is what Erwin and colleagues named their 2011 paper after: the Cambrian conundrum is precisely the mismatch between deep molecular divergence estimates and a shallow fossil record.⁴  Second, the empty Ediacaran preservation windows are a constraint on any deep-root model that requires large soft-bodied animals to have been present.¹³  The standard reply, that the earliest bilaterians were tiny, below the resolution of those windows, is a live hypothesis and is treated as one in Part 2.  Kevin Peterson and colleagues proposed it in 2004 to explain the absence:  "This cryptic precambrian history suggests that these stem-group bilaterians must have been" micrometazoans, animals too small to leave a body fossil.³⁶  No fossil has yet confirmed it.

Third, the clock result is disputed inside the mainstream, by name.  The clock position, in its own words, comes from Mario dos Reis and colleagues in 2015:  their results "suggest that the Cambrian explosion is a phenomenon of fossilization, while biological diversity was established in the Neoproterozoic," with crown animals originating, on the mainstream scale, "833–650 Ma, fully within the Cryogenian."³⁷  Graham Budd and Richard Mann reran that analysis "changing a single number, the soft maximum age of the Metazoa," and found that "the analysis accommodates this shift simply by bunching up the earliest nodes."  Their conclusion is that the clock gives no "evidence for an origin of the clade deeper than what is suggested by the fossil record."³⁸  The clock's authors replied in 2026 with a formal model test and found "strong evidence against the hypothesis that molecular data are uninformative about the age of crown Metazoa," while noting that the comparison is "conditioned on the rate model used."³⁹  So the question is open between specialists, and the fossil side of it rests on exactly this article's Parts 1 and 2:  the trace record and the empty Ediacaran windows.  Both sides read the same radiometric timescale, so the dispute is about whether animals lived long before their fossils, not about the age of the rocks.  The sibling article gives the numbers.

3. "Higher taxa are human categories, so counting phyla is circular."

A fair objection, and this article concedes most of it.  A phylum is a rank a taxonomist assigns, and a distinctive Cambrian animal may get a high rank because it looks unlike anything else, which would make "phyla appear separated by gaps" a restatement of the naming rule.  The series flagship works this through at length, with the mainstream sources that raise it.⁹

Three things blunt it here without removing it.  The Ordovician out-produces the Cambrian at order and family rank while producing one phylum, which is not what a pure naming artefact predicts.  The trace-fossil result uses no Linnaean rank at all: ichnodisparity counts architectural designs.  And this article states its own result as a claim about first appearances as recorded, never as a measurement of morphological distance.

4. "Stem groups fill the gaps, so there are no gaps."

This is the objection that does most damage, and Part 6 states it at full strength rather than answering around it.  Zhang and Shu are explicit that the abundance of stem groups early in animal history is a major reason for the morphological gaps between phyla today, and they name Anomalocaris and Opabinia as forms sitting between three living phylum-rank groups.⁸⁸

The residual question is empirical and this article does not pretend to have answered it: whether a given stem-group fossil is morphologically intermediate or merely basal on a tree has to be settled case by case, from the fossils.  That is what the named-sequence articles in this series are for.

5. "Your own sampling-standardised numbers contradict your own claim."

They do, in part, and both tables are printed.  Once every Cambrian stage is cut to the same number of occurrences, the ratio of classes to genera shows no early peak: the highest value is Stage 4, well after the phylum-rank first appearances have stopped.

What survives is a different quantity.  A first appearance is a one-time event in the record; a standing ratio is what a snapshot of the seafloor would have looked like.  Part 3's result is about the first, and it says so.  Anyone who quotes it as though it were the second is overstating it.

6. "The Sinsk extinction explains why your count stops."

It might, and this article raises the objection before a critic does.  Zhuravlev and Wood place a global anoxic extinction early in Stage 4 that destroys the stem-group skeletal fauna outright.⁵  Part 3's headline result is that phylum-rank first appearances stop at the top of Stage 3, the stage just below.  Those two are close enough that coincidence cannot be assumed.

The test that would settle it has not been run by anyone: a lineage-by-lineage check of whether the groups producing high-rank first appearances are the groups the Sinsk Event killed.  Until that is done, this article's own headline interval carries an asterisk, and the asterisk is printed in Part 3.

7. "This is an argument from ignorance dressed up with numbers."

Taken as an argument for design, it would be, and this article does not make that argument.  Marshall levelled exactly this charge at Meyer, and this site's own rules require taking it seriously rather than deflecting it.³⁴

What this article claims is narrower and checkable.  A rank-structured pattern of first appearances exists and is measured here.  A jump in trace-fossil architecture exists and is measured by ichnologists.  The Ediacaran preservation windows contain algae rather than animals, which is an observation.  And the standard reviews of the subject report that no proposed cause is sufficient to explain the origin of the body plans, which is a description of the literature, not an inference about what caused them.

A reader is entitled to conclude from all of that only that the event is real and poorly explained.  Anyone who wants to go further has to bring an argument this article has not supplied.

8. "Creationists quote scientists out of context. Why should this article be different?"

Because it publishes its own failures first.  This article kills a duration figure its own side has used for thirty years, corrects the citation in the best creationist paper on the subject, cuts an anecdote from its own notes as unsourceable, and prints a one-word misquotation of Gould that this project's own notes carry.

And every quotation here is checked against the author's own next move.  Ohno's short duration runs with Ohno's proposed mechanism, because the mechanism is the reason he cited the duration.  Erwin and Valentine's 1984 sentence about the implausibility of microevolutionary substitution is not quoted anywhere in this article without noting that the paper exists to propose a naturalistic alternative.⁴⁰  Marshall's list of unexplained things runs with Marshall's own expectation that the explanations will be found.  If a future version of this page ever drops one of those pairs, it has been corrupted.

9. "The Ediacaran biota is the ancestral fauna you say is missing."

Partly true, and Part 2 concedes it.  At least one Ediacaran organism is an animal with bilaterian-grade organisation, and it is found with the traces it made.  A second is an animal on chemical evidence.  "Animals appear from nothing" is not a description of this record.

What the objection cannot carry is the rest of the way.  Seven of the nine likely Ediacaran clades cannot be placed in the animal kingdom with confidence.⁴  The biota goes extinct rather than grading into the Cambrian fauna.  And no Ediacaran organism has been identified as the ancestor of any Cambrian phylum, including Kimberella, whose affinity its own describers hedge.

10. "Forty million years is plenty of time for evolution."

Possibly, and this article does not claim otherwise on time alone.  Zhang and Shu, who put the maximum duration at about forty million years, call that rapid in the scheme of Earth history and equal to about 1% of the history of life, and they are arguing for a naturalistic account throughout.⁸

The claim that a duration is too short to accomplish something requires a measured rate for the thing being accomplished, and no such rate exists for the origin of a body plan.  Anyone asserting one is inventing a measurement.  What can be said is what Part 7 says: the specialists themselves report no sufficient cause, and they say so in print.

11. "If a global Flood laid these rocks down, why is there any order at all?"

That is the subject of a separate article in this series, which examines every young-earth explanation of fossil order against an open database and reaches a verdict that goes against the young-earth position.⁴¹  Part 7 states, from their own papers, what young-earth geologists say the Cambrian base itself represents, and where that reading is weak.  Nothing else in the present article depends on the answer.  The Cambrian pattern described here is a statement about what lies above what, and it has to be explained by any model of how the rock was deposited.

What would change this verdict

• A lineage-by-lineage test of the Sinsk confound. If the groups producing phylum-rank first appearances turn out to be the groups the Stage 4 anoxic event destroyed, this article's headline interval is measuring an extinction rather than the end of a radiation. Nobody has run that test, including this article.
• A published partition of the Cambrian signal into evolutionary and taphonomic components. There is no estimate of the form "X% of the apparent rise in first appearances is preservation." Until there is, the magnitude of the explosion is not a measured quantity, and Part 5 says so.
• Ediacaran fossils that are unambiguous members of Cambrian phyla. The Ediacaran preservation windows are the place to look, they have been searched hard, and what they yield is algae. One arthropod from the Miaohe biota would end the argument in this article.
• A morphometric test of whether the named Cambrian stem groups are morphologically intermediate, rather than merely basal on a tree. Radiodonts and lobopodians are the test cases. This decides Part 6 and it has not been done.
• The full text of Mussini and colleagues' 2024 Pikaia paper, which is published under a licence that permits redistribution and which this article could not reach. Its phylogeny is the strongest recent evidence against this article's side, and it is currently assessed here from an abstract.
• The rest of Erwin and Valentine's 2013 monograph. The passage Meyer relies on has now been checked and holds (Part 9), and it strengthens Part 5's second answer without changing the verdict. The book's later chapters on the Ediacaran and on stem groups have not been read for this article.

Sources

The full research dossier for this article, including a quotation ledger with a verdict on every passage used, is published beside it: research/cambrian-fossils.md.  The analysis scripts are research/first-appearances.py and research/rank-first-appearances.py, and the shared query tool is bin/paleofetch.py.